Control and low-level rendering issues in haptic interfaces Tim Salcudean University of British Columbia Department of Electrical and Computer Engineering Vancouver, Canada http://www.ece.ubc.ca/~tims IMA Workshop 9: Haptics, Virtual Reality, and Human Computer Interaction, June 14-15, 2001 REFERENCES WILL BE INSERTED LATER - PLS SEE References.pdf FOR A LIST
32
Embed
Control and low-level rendering issues in haptic interfaces
Control and low-level rendering issues in haptic interfaces. University of British Columbia Department of Electrical and Computer Engineering Vancouver, Canada http://www.ece.ubc.ca/~tims IMA Workshop 9: Haptics, Virtual Reality, and Human Computer Interaction, June 14-15, 2001 - PowerPoint PPT Presentation
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Control and low-level rendering issues in haptic interfaces
Tim SalcudeanUniversity of British Columbia
Department of Electrical and Computer Engineering
Vancouver, Canada
http://www.ece.ubc.ca/~tims
IMA Workshop 9: Haptics, Virtual Reality, and Human Computer Interaction, June 14-15, 2001
REFERENCES WILL BE INSERTED LATER -
PLS SEE References.pdf FOR A LIST
Outline
• Contributors: S.P. DiMaio, K. Hashtrudi-Zaad, M. R. Sirouspour, W.H. Zhu
• List of references will be provided
• introduction and motivation• network model of haptic interaction• stability, performance specifications• controller design approaches and challenges• modelling with hybrid systems• discussion
Introduction/motivation
Some haptic interfaces designed in our lab:• 3-DOF planar device• 5-DOF haptic pen• 6-DOF maglev interface
Introduction/motivationHaptic interface control important:• human hand characteristics difficult to match
– tradeoffs? Performance optimization?
• provides means for psychophysics studies• provides guidelines for design
Studies of:• specific rendering problems• general methodology for coupling user to dynamic
• Four channel architectures allow better tradeoff exploration - Haptic interface behaves as a force or a position sensor
Controller design
• standard linear loop shaping with nominal stability, passivityat the environment• adaptive control• dual hybrid teleoperation• Lyapunov-based methods
• kinematic scaling, impedance shaping possible• velocity control mode possible• force sensing may be replaced by observers
Controller designExperiments with a 3-DOF planar haptic device:
Controller designImpedance simulation: - slave tracks position, force returned
Controller designFully transparent four-channel:- force feedforward, PD controllers for position correspondence
Controller designFour-channel with adaptive damping: - damping proportional to environment force amplitude
Controller designAbsolutely stable four-channel:- reduced force channels to satisfy absolute stability criterion
Adaptive Teleoperation Controller•Adaptive motion/force control of master and slave robots [Zhu ‘97]
,)~
( 2 LLFKXX fd
h, (master)e, (slave)
{
hpfehpee
phd
hfehphped
FKKFAXKXXK
X
FAKXXKXKX
~~~~1
~~~
where
fKpK
w~
-- motion scaling parameter-- force scaling parameter
-- filtered by a first-order filter w
[Anderson 89, Lawrence 92, Yokokohji 92, Colgate 93]
Adaptive Teleoperation Controller
• transparency:
LLXXK
LLXXK
ehp
ehp
2
2
AC
s
K
KZ
K
KZ
f
pe
f
ph
Assuming {e,h} 2nd order LTI systems + the usual on robots, we have:• stability:
Adaptive Teleoperation Controller
Video ...
Adaptive Teleoperation Controller
Hybrid system modelConsider stick-slip friction model:
In practice, implemented as:
Hybrid system model
• video...
Hybrid system model
Hybrid system model
Discussion / Challenges• teleoperation models and controllers applied to haptics• passivity/absolute stability allows modular software• controller analysis is now well understood for linear
and some nonlinear models
• controller synthesis difficult even for simple linear models• adaptive controllers complex and difficult to implement• stability/performance issues not understood for changing
environments
Discussion / Challenges• low level programming is challenging
– need model covering a wide range of behaviours in a (provably?) stable manner